Sunspots and their dynamics dominate the magnetic topology and evolution of both the photosphere and the overlying coronal active regions. Thus a comprehensive understanding of their behaviour is essential to understanding the solar magnetic field. A new technique is presented for applying multiple ellipse fits as a method for rotation tracking of sunspot umbrae. The method is applied to a sunspot in NOAA active region AR 12158 during an X-class flare event and the resulting rotation rate correlates well with other measurements from literature. The method also reveals an apparent difference in rotation rate between the edge and the innermost region of the sunspot umbra of up to 2 degrees per hour. Such differential rotation must lead to the large-scale twisting of sunspot magnetic flux tubes with implications for models of coronal loops and the build-up of instabilities that may lead to eruptions.
Benchmark results for large language models often show inconsistencies across different studies. This paper investigates the challenges of reproducing these results in automatic bugfixing using LLMs, on the HumanEvalFix benchmark. To determine the cause of the differing results in the literature, we attempted to reproduce a subset of them by evaluating 12 models in the DeepSeekCoder, CodeGemma, CodeLlama, and WizardCoder model families, in different sizes and tunings. A total of 35 unique results were reported for these models across studies, of which we successfully reproduced 12. We identified several relevant factors that influenced the results. The base models can be confused with their instruction-tuned variants, making their results better than expected. Incorrect prompt templates or generation length can decrease benchmark performance, as well as using 4-bit quantization. Using sampling instead of greedy decoding can increase the variance, especially with higher temperature values. We found that precision and 8-bit quantization have less influence on benchmark results.
1 January 2016 in Biologically Inspired Cognitive Architectures